Supervised Machine Learning: Regression and Classification
DeepLearning.AI and Stanford Online · Coursera · three modules, taught by Andrew Ng and others
The first course of the Machine Learning Specialization: what machine learning is, linear regression with one input and with many, gradient descent, logistic regression for sorting things into classes, and keeping a model from overfitting. The labs are in Python, with NumPy and scikit-learn in Jupyter notebooks.
Three sessions a module: park a question at the video’s minute, answer it in the lab, and it becomes a recall card.
Module 1 · What machine learning is, with and without labels
Module 1 · A straight-line model, and measuring how wrong it is
Module 1 · Gradient descent: stepping downhill
Module 2 · Many inputs at once, and vectorized code
Module 2 · Scaling features and choosing the learning rate
Module 2 · Making new features, and fitting curves
Module 3 · Sorting into classes with logistic regression
Module 3 · The cost of a classifier, and training it
Module 3 · Overfitting, and holding it back with regularization
Supervised Machine Learning: Regression and Classification is DeepLearning.AI’s and Stanford Online’s, on Coursera. This logbook is Soundings’ own: session names and questions written for it, nothing copied from the course.